Re: [v6ops] RFC4861 question - short prefixes in PIOs

["Pascal Thubert (pthubert)" <[e-mail]>]

Hello Mark

Please keep in mind the use case is wireless (802.11OCB) and the router is a L3 AP. To your point:

On .11 the only direct connectivity that is known is STA to AP. A STA may or may not be reachable from another STA at the PHY layer depending on their relative location. In usual .11 they will not have unicast keys to talk to one another anyway. The L3 AP MUST NOT use redirects.

On OCB in particular there is neither L2 association nor L2 security. The only way for a L3 AP to know who’s there is the L3 association (aka registration, RFC 8505).

A fixed L3 AP (an RSU in OCB) may provide L3 connectivity to vehicles.

If the vehicle moves then it needs heuristics to figure the best router e.g., the most recently seen. When I implemented Mobile IP and NEMO in our routers some 15 years ago I had to place a number of non-standard heuristics in ND to make that work. One of them was to use the router that exposes a PIO as GW for packets that are sourced on the prefix in the PIO.

I started an IPv6 over wireless draft. There’s a lot that needs saying in there.

All the best,

Pascal

Le 29 juin 2019 à 02:50, Mark Smith <[e-mail]<mailto:[e-mail]>> a écrit :



On Sat., 29 Jun. 2019, 02:32 Philip Homburg, <[e-mail]<mailto:[e-mail]>> wrote:
>Hairpin'ing all traffic through the advertising router when L=0 seems like
>the only general solution.  But, I can imagine many situations where if a
>Neighbour Cache entry somehow existed, then it could be used.

Note that the router may send redirects. So in theory it is only the
first packet to a neighbor that needs to go through the router.


You'd have to switch them off in this case,.

That has also been a common thing to do in IPv4 in general. I can't exactly remember why, because it is so long ago that I learned to do it, however I expect it is to stop packets being punted to the control plane for redirect generation. Non-optimal forwarding or dropping, in particular in forwarding hardware, is cheaper than optimal forwarding and control plane load of generating redirects. It also probably creates a control plane DoS attack vector.



--------------------------------------------------------------------
IETF IPv6 working group mailing list
[e-mail]<mailto:[e-mail]>
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[e-mail]<mailto:[e-mail]>
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------